Subjects typically make substantial numbers of errors in visual information processing experiments using speeded presentations. Every theory of visual processing postulates some construct for explaining such errors. These constructs include: failure to extract all of the necessary information from the display; decay of necessary information from some transient store (e.g., iconic store); and various capacity limits (e.g., a limited acoustic, short-time memory). The peturbation model was proposed by the principal investigator to account for error patterns in various visual processing paradigms. It was postulated that feature movement during processing was a primary source of errors. The model was motivated by the failure of existing constructs to provide a natural account of the effects of some salient variables, particularly those relating to lateral interference. Two pilot experiments were carried out using an artificial character set, designed so that feature perturbations, if they occurred, would result in particular error patterns. The pilot results provided some evidence for the concept of feature perturbations. Seven experiments are proposed based on the methodology of the pilot experiments. Their purpose is to verify the pilot results, to test for alternative explanations, and, if verified, to examine the functional properties of the perturbation process.